An earlier commit fixed an issue in "check_describe" with
"test_expect_success" being called within another
"test_expect_success", causing the test to succeed even if it should
fail.
Let's try to guard against this in the test library by returning 1
from these two functions. This change would have caught the issue I've
now fixed in the "check_describe" function.
I could equivalently add this "return 1" to the "test_finish_"
function itself, but I think doing it here is more readable.
Because of this change any tests which ran under "set -e" needed to be
refactored not to use "set -e". Luckily there were only two such
tests, earlier commits did that refactoring.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@xxxxxxxxx>
---
t/test-lib-functions.sh | 2 ++
1 file changed, 2 insertions(+)
diff --git a/t/test-lib-functions.sh b/t/test-lib-functions.sh
index c6cdabf53e..3dd68091bb 100644
--- a/t/test-lib-functions.sh
+++ b/t/test-lib-functions.sh
@@ -636,6 +636,7 @@ test_expect_failure () {
fi
fi
test_finish_
+ return 1
}
test_expect_success () {
@@ -656,6 +657,7 @@ test_expect_success () {
fi
fi
test_finish_
+ return 1
}
# test_external runs external test scripts that provide continuous
--
2.31.0.rc0.116.g45ec00aa00
